On the Description box of the image generator there is a little i which says “Add brackets around text (like this) or ((even more)) to get the AI to pay more attention to it”. Is this limited to 2 sets of brackets or does it continue, that having (((something))) makes the AI pay even more attention and then 4 sets even more and so on?
Wait, are you saying everything here https://perchance.org/tutorial is not used anymore?
That was written for Stable Diffusion. The new generator is Flux.1-Schnell based.
Ah right, got you, thanks.
WE NOW USE FLUX? When did that happen?
Hm, it says SD.
2025/05/17
All of it is out of date. Perchance switched to Flux awhile ago. There has been multiple posts about the switch and the dev had a a message they put when they switched.
No wonder the quality of the images changed.
Oh, so the brackets, including one or two, do nothing. Right, got you, thanks.
FLUX.1 uses natural language. Write full sentences as if describing the image to a human artist. Avoid tag-style keywords, brackets, or weight syntax like (tag:1.3). Avoid Stable Diffusion-style prompt features such as ((tag)), (tag:1.3), BREAK, AND, or OR. These are not supported and will be ignored by FLUX.1.
I believe it’s kinda outdated. The new model prefers full sentences with details, rather than a word with brackets.
